TabPFN EduLab - Interactive Learning

Dual-Attention X-ray

TabPFN reads tables with two different attention mechanisms using transformer architecture. into a cell from the table below by clicking Examine how the model “sees” the painting at that moment.

Attention Mechanism:

Patient ID	Age	Tansiyon	Kolesterol	Disease Risk (Target)

Please select a cell from the table

Click any attribute (Age, Blood Pressure, etc.) in the table above to see how the model establishes relationships within the data.

Time Series Converter (TabPFN-TS)

TabPFN-TS uses time series forecasting instead of a complex network architecture. to a tabular regression problem transforms. See how this transformation works by plotting your own signal.

1. Plot Your Signal

Draw a wave inside the box with the mouse or by touching it.

Feature Extraction Formula

D = 18 + 2k + D_z

Each time step is converted into a row of at least 28 columns (D) for the model to understand.

Bekleniyor...

2. Data Structure Seen by TabPFN

Time (t)	Hedef (Y)	Running Index	Takvim: sin(t)	Takvim: cos(t)	...	Kovaryat (Dz)

Injective Embedding: the model does not use sin(t) alone. Using sine alone could confuse equal values on the rising and falling parts of a wave. Using both sine and cosine maps time onto the unit circle while preserving phase information.